Functions — github.com/Lancelot39/KGSF

CodeHub MCPcopy Create free account

hub / github.com/Lancelot39/KGSF / functions

Functions130 in github.com/Lancelot39/KGSF

⨍ Functions 130◇ Types & classes 31

↓ 13 callersMethod__init__

(self, dim=1, attn='cosine')

models/transformer.py:126

↓ 10 callersFunction_normalize

Broadcast layer norm

models/transformer.py:18

↓ 9 callersMethod__init__

(self, ninp, nhid, dropout=0.5)

models/graph.py:59

↓ 5 callersFunctioncreate_position_codes

(n_pos, dim, out)

models/transformer.py:114

↓ 5 callersMethoddata_process

(self,is_finetune=False)

↓ 5 callersMethodpadding_w2v

(self,sentence,max_length,transformer=True,pad=0,end=2,unk=3)

↓ 3 callersFunction_build_encoder

(opt, dictionary, embedding=None, padding_idx=None, reduction=True, n_positions=1024)

models/transformer.py:23

↓ 3 callersMethodbackward

Perform a backward pass. It is recommended you use this instead of loss.backward(), for integration with distributed training and F

↓ 3 callersMethodtrain

↓ 3 callersMethodval

(self,is_test=False)

↓ 3 callersMethodvector2sentence

(self,batch_sen)

↓ 3 callersMethodzero_grad

Zero out optimizer. It is recommended you call this in train_step. It automatically handles gradient accumulation if age

↓ 2 callersFunction_create_embeddings

Create and initialize word embeddings.

models/utils.py:31

↓ 2 callersMethod_starts

Return bsz start tokens.

↓ 2 callersMethod_starts

Return bsz start tokens.

models/transformer.py:1132

↓ 2 callersMethodload_model

↓ 2 callersMethodoutput

↓ 2 callersMethodsave_model

↓ 2 callersMethodupdate_params

Perform step of optimization, clipping gradients and adjusting LR schedule if needed. Gradient accumulation is also performed if ag

↓ 2 callersMethodval

(self,is_test=False)

↓ 1 callersFunction_build_decoder4kg

(opt, dictionary, embedding=None, padding_idx=None, n_positions=1024)

models/transformer.py:96

↓ 1 callersMethod_context_reformulate

(self,context,movies,altitude,ini_altitude,s_id,re_id)

↓ 1 callersFunction_create_entity_embeddings

Create and initialize word embeddings.

models/utils.py:42

↓ 1 callersMethod_create_selfattn_mask

models/transformer.py:701

↓ 1 callersMethod_create_selfattn_mask

models/transformer.py:893

↓ 1 callersFunction_edge_list

(kg, n_entity, hop)

↓ 1 callersMethodbackward

Perform a backward pass. It is recommended you use this instead of loss.backward(), for integration with distributed training and F

↓ 1 callersMethodbuild_model

(self,is_finetune)

↓ 1 callersMethodbuild_model

(self,is_finetune)

↓ 1 callersMethodcompute_loss

(self, output, scores)

↓ 1 callersFunctionconcept_edge_list4GCN

↓ 1 callersMethoddecode_forced

Decode with a fixed, true sequence, computing loss. Useful for training, or ranking fixed candidates. :param ys:

↓ 1 callersMethoddecode_forced

Decode with a fixed, true sequence, computing loss. Useful for training, or ranking fixed candidates. :param ys:

models/transformer.py:1177

↓ 1 callersMethoddecode_greedy

Greedy search :param int bsz: Batch size. Because encoder_states is model-specific, it cannot infer thi

↓ 1 callersMethoddecode_greedy

Greedy search :param int bsz: Batch size. Because encoder_states is model-specific, it cannot infer thi

models/transformer.py:1136

↓ 1 callersMethoddetect_movie

(self,sentence,movies)

↓ 1 callersMethodencode_cand

models/transformer.py:1054

↓ 1 callersMethodencode_context_memory

(self, context_w, memories_w)

models/transformer.py:1072

↓ 1 callersMethodinfomax_loss

(self, con_nodes_features, db_nodes_features, con_user_emb, db_user_emb, con_label, db_label, mask)

↓ 1 callersMethodinit_optim

Initialize optimizer with model parameters. :param params: parameters from the model :param optim_states:

↓ 1 callersMethodinit_optim

Initialize optimizer with model parameters. :param params: parameters from the model :param optim_states:

↓ 1 callersFunctionkaiming_reset_parameters

(linear_module)

models/graph.py:14

↓ 1 callersMethodmetrics_cal_gen

(self,rec_loss,preds,responses,recs)

↓ 1 callersMethodmetrics_cal_rec

(self,rec_loss,scores,labels)

↓ 1 callersFunctionneginf

Returns a representable finite number near -inf for a dtype.

models/utils.py:24

↓ 1 callersMethodoptim_opts

Fetch optimizer selection. By default, collects everything in torch.optim, as well as importing: - qhm / qhmadam if inst

↓ 1 callersMethodoptim_opts

Fetch optimizer selection. By default, collects everything in torch.optim, as well as importing: - qhm / qhmadam if inst

↓ 1 callersMethodpadding_context

(self,contexts,pad=0,transformer=True)

↓ 1 callersMethodreset_parameters

models/graph.py:37

↓ 1 callersMethodresponse_delibration

(self,response,unk='MASKED_WORD')

↓ 1 callersFunctionsetup_args

↓ 1 callersMethodupdate_params

Perform step of optimization, clipping gradients and adjusting LR schedule if needed. Gradient accumulation is also performed if ag

↓ 1 callersMethodzero_grad

Zero out optimizer. It is recommended you call this in train_step. It automatically handles gradient accumulation if age

Method__getitem__

movie_vec = np.zeros(self.entity_num, dtype=np.float) context, c_lengths, response, r_length, entity, movie, concept_mask, dbpedia_

(self, opt, is_finetune)

(self, opt, is_finetune)

(self,filename,opt)

(self, dataset, entity_num, concept_num)

(self, opt, dictionary, is_finetune=False, padding_idx=0, start_idx=1, end_idx=2, longest_label=1)

(self, in_features, out_features, bias=True)

models/graph.py:26

(self, in_features, out_features, dropout, alpha, concat=True)

models/graph.py:79

(self, dim, da, alpha=0.2, dropout=0.5)

models/graph.py:116

(self, dim, da, alpha=0.2, dropout=0.5)

models/graph.py:142

(self, dim, da)

models/graph.py:172

(self, input_size, dropout)

models/graph.py:193

Dense version of GAT.

models/graph.py:221

(self, in_features, out_features, dropout, alpha, concat=True)

models/graph.py:270

Sparse version of GAT.

models/graph.py:331

(self, n_heads, dim, dropout=0)

models/transformer.py:150

(self, dim, dim_hidden, relu_dropout=0)

models/transformer.py:233

(self, transformer, hdim)

models/transformer.py:250

( self, n_heads, n_layers, embedding_size, ffn_size, dro

models/transformer.py:289

( self, n_heads, embedding_size, ffn_size, attention_dropout=0.0,

models/transformer.py:367

( self, n_heads, n_layers, embedding_size, ffn_size, voc

models/transformer.py:422

( self, n_heads, n_layers, embedding_size, ffn_size, voc

models/transformer.py:544

( self, n_heads, embedding_size, ffn_size, attention_dropout=0.0,

models/transformer.py:644

( self, n_heads, n_layers, embedding_size, ffn_size, voc

models/transformer.py:736

( self, n_heads, embedding_size, ffn_size, attention_dropout=0.0,

models/transformer.py:804

( self, n_heads, n_layers, embedding_size, ffn_size, voc

models/transformer.py:928

(self, opt, dictionary)

models/transformer.py:999

( self, padding_idx=0, start_idx=1, end_idx=2, unknown_idx=3,

models/transformer.py:1117

models/graph.py:53

models/graph.py:112

models/graph.py:327

Function_add_neighbors

(kg, g, seed_set, hop)

models/graph.py:364

Function_build_decoder

(opt, dictionary, embedding=None, padding_idx=None, n_positions=1024)

models/transformer.py:78

Function_build_encoder4kg

(opt, padding_idx=None, reduction=True, n_positions=1024)

models/transformer.py:42

Function_build_encoder_mask

(opt, dictionary, embedding=None, padding_idx=None, reduction=True, n_positions=1024)

models/transformer.py:59

Function_load_kg_embeddings

(entity2entityId, dim, embedding_path)

(ctx, grad_output)

models/graph.py:250

Methodco_occurance_ext

Methoddistinct_metrics

Methodentities2ids

(self,entities)

Get output predictions from the model. :param xs: input to the encoder :type xs: LongTensor[bs

(self, input, adj)

models/graph.py:45

x: shape (|V|, |D|); adj: shape(|V|, |V|)

models/graph.py:66

(self, input, adj)

models/graph.py:94

models/graph.py:130

next →1–100 of 130, ranked by callers